A new amyloid PET evaluation method using separated gray-matter histogram based on three-component model.

OBJECTIVE: We have constructed a three-component model underlying amyloid PET accumulation and developed a new gray matter histogram evaluation method based on this model. This study aims to validate the utility of the new method compared with conventional visual and SUVR-based quantitative evaluation. METHODS: A retrospective analysis was performed on amyloid PET/CT data from 63 participants (25 healthy volunteers, 38 patients with dementia or cognitive impairment) of previous study using 18F-FPYBF-2. Subjects were visually classified into three groups: negative, borderline, and positive, and quantitatively evaluated using composed standardized uptake value (comSUVR) with a reference to cerebellar cortex. Histograms were generated for the whole-brain, gray matter (GM-histogram), and white matter (WM-histogram) based on probability-tissue maps. The GM-histogram was further decomposed into two Gaussian components: G1 and G2　using statistical software. Parameters of whole-brain histogram: skewness, mode-to-mean ratio (MMR), and parameters of GM-histogram: GM-kurtosis, µG2 (mean of G2), and πG2 (proportion of G2), were compared among visual groups and the correlation with comSUVR was evaluated. RESULTS: The GM-histogram was sharply unimodal in visually negative group but showed a wide shape to bimodal patterns in visually positive cases. Visually border group showed significantly higher πG2 than negative group, and positive group showed significantly higher µG2 than border group. GM-kurtosis and µG2 showed stronger negative (p < 0.0001, R2 = 0.7539) and positive (p < 0.0001, R2 = 0.8589) correlations with ComSUVR, respectively than the correlations between whole-brain histogram parameters and ComSUVR. CONCLUSION: Our proposed GM-histogram provides a visually comprehensive morphology and quantitative indicators that match conventional visual and SUVR-based assessments and may potentially detect even subtle amyloid accumulation. This method is considered promising as a complementary tool for early diagnosis and treatment monitoring of Alzheimer’s disease.